2022-2023
- Work in groups of 2, 3 or 4 students, assigned at random (list on the Moodle website).
Instructions
The project will be two-fold: handle datasets from French electricity consumption with the motivation to create
- a visualization inspired by the image below,
- to predict electricity consumption on a future day.
Data link for prediction: for this part the dataset is available at https://odre.opendatasoft.com/explore/dataset/eco2mix-national-tr/information/?disjunctive.nature&sort=-date_heure. The goal is simply to predict the consumption on Thursday 8 December for the whole day, at the country-wise scale. We will measure your score as a Mean Square Error on the 96 quarters of hours during that day, so your output should simply be a .csv file with the following structure:
| Date | Heure | Consommation (MW) |
|---|---|---|
| 2022-12-08 | 00:00 | 30811 |
| 2022-12-08 | 00:15 | 30889 |
| … | … | … |
| 2022-12-08 | 23:45 | 30878 |
You will also display your prediction for all the sources (Gas, wind, etc.) on a simple graph of your choice.
Data link for visualization: For this part, the dataset is available at https://data.enedis.fr/explore/dataset/consommation-annuelle-residentielle-par-adresse/information/.
The motivation is to analyze and understand better the electricity consumption of the French household.
The main task for this part of the project is to provide an interactive map of France at the city-wise level displaying the yearly average household electricity consumption between 2018 and 2021. The map should be clickable, and when clicking on a city, you would display the distribution (with a violin plot, a swarmplot or a KDE plot for instance) of consumption per household in this city over the four years.
Hint: you could be inspired by this gif